
cifar-10-batches-py.zip
5星
- 浏览量: 0
- 大小:None
- 文件类型:ZIP
简介:
在完成cs231n课程作业时所使用的机器学习分类数据集。由于国内下载速度极度缓慢,并且运行该脚本需要依赖Linux系统,因此我决定直接将相关内容发布到平台。
CIFAR-10和CIFAR-100是800万小图像数据集中的标记子集。这些图像由Alex Krizhevsky、Vinod Nair和Geoffrey Hinton收集。
CIFAR-10数据集包含60000张32x32像素彩色图像,分为10个类别,每个类别包含6000张图像。其中,有50000张用于训练,10000张用于测试。
The dataset is structured into five distinct training sets and a single evaluation set, each comprising 10,000 images. The evaluation set precisely includes 1,000 randomly chosen images representing every class. The training sets encompass the remaining images, arranged in a random sequence; however, it’s possible for certain training sets to feature a disproportionate number of images belonging to particular classes compared to others. Collectively, these training sets contain precisely 5,000 images per class.
全部评论 (0)


